Pre-gated MoE: An Algorithm-System Co-Design for Fast and Scalable Mixture-of-Expert Inference.
Ranggi HwangJianyu WeiShijie CaoChangho HwangXiaohu TangTing CaoMao YangMinsoo RhuPublished in: CoRR (2023)
Keyphrases
- detection algorithm
- learning algorithm
- significant improvement
- cost function
- preprocessing
- computational cost
- optimization algorithm
- times faster
- experimental evaluation
- worst case
- matching algorithm
- segmentation algorithm
- expectation maximization
- inference process
- theoretical analysis
- single pass
- improved algorithm
- np hard
- loopy belief propagation
- data sets
- classification algorithm
- input data
- denoising
- neural network